To determine the differences between a 3D site search of our library with different options, and how they differ from Find-A-Drug's preliminary results. The results will be a preliminary measure of determining our synthetic targets.
Procedure
3D searches were performed on this library, simultaneously, on a lab computer using same settings that produced what we thought were the be results. The procedure followed is as shown in D-EXP002. The protein and site used were the malaria enoyl reductase protein (1p45-q1.pdb). The settings were as follows:
4 centers, tolerance 0.5
4 centers, tolerance 1
3 centers, tolerance 0.5
3 centers, tolerance 1
All were run with a warp speed set as 1.
All were allowed to run to completion, with the best scores (lowest free energies) to be considered for synthetic targets.
Results
Combined Data - All four runs, including Find-A-Drug's data, with sorting filters included. Blank fields mean that the molecule did not dock.
Below are the .sdf files of the docked molecules 4 centers, tolerance 0.5, runtime 12h, 58m (200 out of 218 docked) 4 centers, tolerance 1, runtime 7d, 12h, 40m (218 of 218 docked) 3 centers, tolerance 0.5, runtime 6d, 21h, 49m (218 of 218 docked) 3 centers, tolerance 1, runtime 7d, 8h, 45m (218 of 218 docked) Find-A-Drug's results (SMILES) [note: the labels within the brackets ( { } ) cannot be read by some programs. The brackets can be removed and then can be read with no problem, and can also be handled in THINK itself. We're in the process of determining the nature of the labels, how they got there, and what they mean. The data is listed next to each of the SMILES codes.]
None of our results in the combined data file replicated Find-A-Drug results, either in absolute values or relative rank. The long runtimes can be explained due to the fact that they were all large libraries run simultaneously on the same computer, slowing down the processes of each. Run individually, they would probably be done in under 3 days on most computers. Ideally, running them on the same computer would yield the same computational round-off errors, which can vary between processors of different computers.
Conclusion
We cannot reproduce Find-A-Drug's THINK results using 3 or 4 centers and 0.5 or 1 tolerance.
Objective
To determine the differences between a 3D site search of our library with different options, and how they differ from Find-A-Drug's preliminary results. The results will be a preliminary measure of determining our synthetic targets.Procedure
3D searches were performed on this library, simultaneously, on a lab computer using same settings that produced what we thought were the be results. The procedure followed is as shown in D-EXP002. The protein and site used were the malaria enoyl reductase protein (1p45-q1.pdb). The settings were as follows:4 centers, tolerance 0.5
4 centers, tolerance 1
3 centers, tolerance 0.5
3 centers, tolerance 1
All were run with a warp speed set as 1.
All were allowed to run to completion, with the best scores (lowest free energies) to be considered for synthetic targets.
Results
Combined Data - All four runs, including Find-A-Drug's data, with sorting filters included. Blank fields mean that the molecule did not dock.
Below are the .sdf files of the docked molecules
4 centers, tolerance 0.5, runtime 12h, 58m (200 out of 218 docked)
4 centers, tolerance 1, runtime 7d, 12h, 40m (218 of 218 docked)
3 centers, tolerance 0.5, runtime 6d, 21h, 49m (218 of 218 docked)
3 centers, tolerance 1, runtime 7d, 8h, 45m (218 of 218 docked)
Find-A-Drug's results (SMILES) [note: the labels within the brackets ( { } ) cannot be read by some programs. The brackets can be removed and then can be read with no problem, and can also be handled in THINK itself. We're in the process of determining the nature of the labels, how they got there, and what they mean. The data is listed next to each of the SMILES codes.]
Below are .xls files, using Open Babel to convert from SDF to SMILES.
4 centers, tolerance 0.5
4 centers, tolerance 1
3 centers, tolerance 0.5
3 centers, tolerance 1
Discussion
None of our results in the combined data file replicated Find-A-Drug results, either in absolute values or relative rank. The long runtimes can be explained due to the fact that they were all large libraries run simultaneously on the same computer, slowing down the processes of each. Run individually, they would probably be done in under 3 days on most computers. Ideally, running them on the same computer would yield the same computational round-off errors, which can vary between processors of different computers.Conclusion
We cannot reproduce Find-A-Drug's THINK results using 3 or 4 centers and 0.5 or 1 tolerance.